How to Turn a Long Customer Review Into a Five-Bullet Case Study

Anyone sitting on a long customer review can turn it into a five-bullet case study in 12 minutes: the problem, one quote, one metric, the result, and what a reader should do next. Run it in ChatGPT, Claude, or Gemini before a "heartwarming success story" invents a brochure.

A five-bullet case study is proof, not praise. If the page still reads like a thank-you email, you never briefed a seller. Five labeled bullets tell one reader what changed. A wall of adjectives tells no one.

This is not a product description that is not a brochure. A description names what you sell. This skill names what one buyer already proved. It is not captions that do not sound like ads. A caption is a scroll stop. A case study is a sales leave-behind. It is not a rambling status update into three lines. A status is mid-week work. This is after a result landed. Save the five-bullet shape as a skill prompt you can rerun.

Build a five-bullet case study from one review

1. Paste the review. Do not draft a brochure

Instruct ChatGPT, Claude, or Gemini: "Do not draft a case study yet. Paste back the review I give you. If a sentence has no customer fact, write UNKNOWN on that sentence. No cleanup. No brand voice. No we are thrilled."

Paste the App Store review, the G2 dump, the email that said "this finally fixed," or the Slack thread they typed after go-live. A model that "upgrades" their nouns into your homepage is lying on your behalf.

Write five facts on paper before the bullets: who they are in one noun (role or company type), what was broken, whether a number appears, whether a before state is clear, and whether they named a next step or you must leave it blank.

Banned: thrilled to partner, delighted customer, transformative journey, seamless experience, game changer, best-in-class, end-to-end, leveraging our platform, excited to share, love letter to the brand, success story that writes itself.

2. Name the problem in their words, then freeze one quote

Send the review back. Instruct: "Bullet one is Problem: one sentence in the customer's nouns for what was broken. Bullet two is Quote: one short line they actually typed, in quotation marks. If I named no clear problem, write UNKNOWN and stop. Do not invent a persona."

Good problem: "Problem: Friday vendor CSV still needed a 40-minute hand fix before finance would touch it." Bad: "Problem: they struggled with inefficient workflows." Workflows is your brochure.

Good quote: "Quote: 'I stopped babysitting the tax column after week two.'" Bad: "Quote: 'Amazing product, highly recommend!!!'" That is a star rating, not proof.

If two problems appear, keep the one with a clock or a cost. Put the second on a second card. Two problems in one five-bullet case study is a paragraph with labels.

3. Metric, Result, Next. UNKNOWN is a legal word

Instruct: "Bullet three is Metric: one number from the paste, or UNKNOWN. Bullet four is Result: what stopped or started after the fix, in one sentence. Bullet five is Next: one verb for the reader (book a demo, try the template, open the guide). If I named no next, write See the product page. Do not invent 3x. Do not invent a CFO title."

Good metric: "Metric: hand-fix time dropped from 40 minutes to 6." Bad: "Metric: massive efficiency gains." Gains are a vibe.

Good result: "Result: finance pulls the file Tuesday without a Slack chase." Bad: "Result: the team feels empowered." Empowered is a poster.

Good next: "Next: paste your own review into the same five bullets." Bad: "Next: join our community of innovators." Community is a funnel you just glued on.

If there is no number, write UNKNOWN on Metric. Do not invent 40% so the card looks finished. A fake percentage is a case study you cannot defend.

4. Force five bullets, then stop the model

Hard cap: five bullets. No hero paragraph above. No testimonial novel under Quote. No logo wall.

Tell the model: "Output exactly five lines labeled Problem, Quote, Metric, Result, Next. No sixth line. No We are proud. No customer journey map. If a banned phrase appears, rewrite that line only."

ChatGPT will add a warm intro and a "hope this helps sellers." Claude will hedge the metric into "roughly faster." Gemini will invent a company size so the card looks complete. Cut all three.

If a sixth line appears, delete it. If Quote is three paragraphs because the model wanted to be thorough, you failed the job. Thorough is the review you already had.

5. One human pass as a cold reader who never met the customer

Read the five bullets as if you will paste them into a sales deck in the next hour.

Check three traps:

• A problem that is actually a feature list. "Needed better automation" is not a problem. "Friday CSV needed a 40-minute hand fix" is.
• A metric the review never said. If you cannot point at the number in the paste, write UNKNOWN.
• A next that sells a webinar the review never implied.

Fix those in your own words from the review. Then send the edited five back: "Replace only the words I changed. Do not re-expand. Do not add a brochure."

One pass. Paste into the deck, the landing page, or the CRM note. If a paragraph still sits above the five bullets, you never wrote a case study. Stop.

Reusable skill card

Trigger: You are about to paste a long review and type "turn this into a case study" or "write a success story."

Input: The review, the customer's role noun if it exists, one number if it exists, plus the banned-phrase list.

Output: Five lines labeled Problem, Quote, Metric, Result, Next. No greeting. UNKNOWN where a field was not in the paste.

Stop: After one human pass as a cold reader. No sixth line. No invented percentage. No second problem hiding in Result. No full review under the bullets.

Takeaways

• Five bullets: problem, quote, metric, result, next. Then stop.
• A review is source. A five-bullet case study is proof. Do not ship both.
• UNKNOWN on the metric is better than a fake 3x. Their nouns beat your brochure.
• Two problems is two cards. A hedge is Unknown, not a pick.